A Simplified Jackknife Approach for the Parameter Uncertainty Analysis of an Urban-Specific Rainfall-Runoff Model

Predictions made by the rainfall-runoff models are inherently uncertain in nature and it is very vital that these models should undergo vigorous calibration and uncertainty analysis. Recent researches relating to hydrologic model uncertainty mostly refers to the identification of parameter uncertainty. The quantitative evaluation of parameter uncertainty of rainfall-runoff models is very important especially in urban watersheds due to the high flood risk in these areas. Therefore, this study aims to analyze the parameter uncertainty of an urban specific rainfall-runoff model, urban storage function (USF) model, using the simplified jackknife approach and its effect on the model simulations. The use of jackknife procedure, a resampling technique, to assess the parameter uncertainty of rainfall-runoff models appear not to have been tried before. The standard rainfall-runoff-model calibration procedure is applied by treating as missing each block of the model residuals to the objective function with a block length of 50. In this study, we scrupulously evaluated the uncertainty of USF model parameters by estimating the 95% confidence interval (CI) of parameters and identified the parameters from the highest to the least uncertainties. Further, the effect of parameter uncertainty on the model simulation uncertainty was investigated by computing 95% CI of the simulated discharge series. The results revealed that the model was able to bracket only 43% of the observations, on average, within the confidence band which further disclosed that the parameter uncertainty has a great impact on the USF model simulation uncertainty.